document resume author mcmillan, james h.: … · reactions of college students in a real-life...
TRANSCRIPT
DOCUMENT RESUME
ED 196 950 TM 810 091
AUTHOR McMillan, James H.: Sprat, Kevin F.TITLE Causal Attributions and Affect in a Real -life Testing
Situation.PUB DATE 80NOTE 27p.: Paper presented at the Annual Meeting of the
American Educational Research Association (64th,Boston, MA, April 7-11, 19B0).
UPS PRICEEESCRIPTORS
MF01/PCO2 Plus Postage.Academic Achievement: Academic Failure; *AffectiveBehavior: Attitude Measures: *Attribution Theory;*College Students: *Factor Analysis; HigherEducation: *Testing
ABSTRACTThis study investigated the nature of affective
reactions of college students in a real-life testing situation, andthe relationship of these affects to causal attributions andperceptions of success or failure. Ninety students responded toquestions concerning their emotions after receiving results of anexamination and indicated the contribution of four causal factors(effort, ability, task difficulty, and luck) as reasons for theirsuccess or failure. Principal component analyses of 14 affectsresulted in two factors: (1) general happiness, pride, andsatisfaction: and (2) surprise. Regression analyses indicated thatperception of success cr failure accounted for a very high proportionof the variance, although internal causal attributions were relatedto positive affect. A procedure for analyzing attributions aspatterns rather than singular factors is presented. (Author)
***********************************************************************Reproductions supplied by EDRS are the best that can be made
from the original document.***********************************************************************
U S DEPARTMENT OF HEALTH.EDUCATION & WELFARE
C=3NATIONAL INSTITUTE OF
EDUCATION
Ln THIS DOCUMENT HAS BEEN REPRO.DUCED EXACTLY AS RECEIVED FROM
Ch THE PERSON OR ORGANIZATION ORIGIN.AT iNG IT POINTS OF VIEW OR OPINIONS
%C STATED 00 NOT NECESSARILY REPRESENT OFFICIAL NATIONAL INSTITUTE OF
CN EDUCATION POSITION OR POLICY
Causal Attributions and Affect
in a Real-life Testing Situation1
by
James H. McMillanVirginia Commonwealth University
Kevin F. SprattThe University of Iowa
"PERMISSION TO REPRODUCE THISMATERIAL HAS BEEN GRANTED BY
J. f(, Mc //gun
TO THE EDUCATIONAL RESOURCESINFORMATION CENTER (ERIC)."
Paper presented at the 1980 Annual Convention of the American EducationalResearch Assosiation, Boston, April 11.
1 The authors would like to thank Margaret M. Clifford and Robert A.Forsyth for their helpful comments and suggestions on an earlier draftof the manuscript.
tlO
O60
Causal Attributions and Affect
in a Real-life Testing Situation
Abstract
This study investigated the nature of affective reactions of college
students in a real-life testing situation, and the relationship of
these affects to causal attributionsand perceptions of success or
failure. Ninety students responded to questions concerning their
emotions after receiving results of an examination and indicated the
contribution of four causal factors(effort, ability, task difficulty,
and luck) as reasons for their success or failure. Principal compo-
nent analyses of 14 affects resulted in two factors, 1) general happiness,pride, and satisfaction; and 2) surprise. Regression analyses indicatedthat perception of success or failure accounted for a very high proportionof the variance, although internal causal attributions were related to
positive affect. A procedure for analyzing attributions as patterns
rather than singular factors is presented.
Causal Attributions and Affect
in a Real-life Testing Situation)
by
James H. McMillanVirginia Commonwealth University
Kevin F. SpratThe University of Iowa
Running Title: Attributions and Affect
1The authors would like to thank Margaret M. Clifford and Robert A. Forsyth
for their helpful comments and suggestions on an earlier draft of the
manuscript.
4
Weiner and his colleagues (Weiner, 1977, 1979; Weiner, Frieze, Kukla,
Reed, Rest, and Rosenbaum, 1971) have presented a cognitive interpretation
of achievement motivation which has generated a substantial amount of research.
Based on Heider (1958), Weiner and others have investigated the extent to
which perceptions of causality influence expectancies of future performance,
affective reactions, and eventually subsequent performance. Although a
large number of research investigations support the theoretical contentions,
there have been few studies which have examined the theory in the context of
real-life achievement situations (Bar-Tal & Darom, 1979; Weiner, 1979).
The present experiment investigated one aspect of the Weiner model, the re-
lationship between perceptions of causality to affective reactions toward an
achievement outcome, in a real-life testing situation.
Previous research has indicated that different attributions tend to
foster different affects or emotions, but the studies have limited general-
izability and offer contradictory findings. In the first set of experiments
in the area (Weiner & Kukla, 1970; Weiner, Frieze, Kukla, Reed, Rest, &
Rosenbaum, 1971) the general conclusion was that internal attributions
(effort, ability) generate more intense affective reactions than external
attributions (task difficulty, luck). The measrre of "affect" in these
experiments was a pride-shame dimension, and it was found that greatest
pride for success and shame for failure was reported when the subject at-
tributed the cause to ability or effort, rather than task difficulty or
luck. The experiments also indicated that effort ascriptions generated
more pride than ability ascriptions. Additional research, utilizing the
same experimental paradigm, has replicated these findings (Reimer, 1975;
Weiner, 1974; Weiner & Kukla, 1970), but the nature of the studies limit
the theoretical utility of the results. First, these studies have typically
2
used contrived situations in which subjects projected the feelings of
others after reading a short paragraph. Second, only one causal attri-
bution was used in each paragraph. Third, tit- measure of affect was
limited to a singe dimension of pride-shame.
Research dome in more real-life like settings has resulted in
contradictory findings (Arkin and Maruyama, 1979; Bailey, Helm and
Gladstone, 1975; Covington and Omelich, 1979; Frieze, Snyder, and
Fontaine, 1977; Nicholls, 1976; and Sohn, 1977). Bailey et. al.,
Frieze et. al., and Arkin and Maruyama assessed attributions and affect
of college students after taking an actual examination in a college
course. All three studies found that outcome (success or failure) was
thy: primary determinant of affect, regardless of whether the internal
ascription was effort or ability. Bailey et. al. found that task ease
generated as much positive affect as internal factors, although this re-
sult could have been a function of the measure of "affect," since several
items (good-bad,satisfied-dissatisfied, ashamed-proud, and happy-sad)
were combined to assess affect rather than using a single pride-shame
dimension. Frieze et. al. also found positive affect (pride-shame) re-
lated more to ability, not effort, and that success and failure were more
highly related to actual performance than internal or external ascriptions.
Arkin and Maruyama, assessing college students' reactions to a course and
instructor, used test anxiety as a measure of affect. They found course
and instructor ratings of successful students positively related to intern-
ality of attributions (the more internal, the more positive), while anxietyof unsuccessful students was negatively related to internality. Nicholls,by asking college students under what conditions they would feel greatestpride or shame, found that effort was more important in determining affect
used contrived situations in which subjects projected the feelings of
others after reading a short paragraph. Second, only one causal attri-
bution was used in each paragraph. Third, the measure of affect was
limited to a single dimension of pride-shame.
Research done in more real-life like settings has resulted
contradictory findings (Arkin and Maruyama, 1979; Bailey, Helm and
Gladstone, 1975; Covington and Omelich, 1979; Frieze, Snyder, and
Fontaine, 1977; Nicholls, 1976; and Sohn, 1977). Bailey et. al.,
Frieze et. al., and Arkin and Maruyama assessed attributions and affect
of college students after taking an actual examination in a college
course. All three studies found that outcome (success or failure) was
the primary determinant of affect, regardless of whether the internal
ascription was effort or ability. Bailey et. al. found that task ease
generated as much positive affect as internal factors, although this re-
sult could have been a function of the measure of "affect," since several
items (good-bad, satisfied-dissatisfied, ashamed-proud, and hopy-sad)
were combined to assess affect rather than using a single pride.-shame
dimension. Frieze et. al. also found positive affect (pride-shame) re-
lated more to ability, not effort, and that success and fa:lure were more
highly related to actual performance than internal or external ascriptions.
Arkin and Maruyama, assessing college students' reactions to a course and
instructor, used test anxiety as a measure of affect. They found course
and instructor ratings of successful students positively related to intern-
ality of attributions (the more internal, the more positive), while anxiety
of unsuccessful students was negatively related to internality. Nicholls,
by asking college students under what Jnditions they would feel greatest
pride or shame, found that effort was more important in d, f..ermining affect
7
3
for courses with no long-term consequences, but that ability was valued
more if the performance Was related to long terra goals. Using a similar
procedure, Sohn showed that happiness is equally influenced by an ability
as by an effort ascription,whereas pride-shame is affected most by effort
ascriptions. Covington and Omelich, using path analysis techniques of
college students' attributions and affect (pride-shame) following failure
on an actual psychologyexamination, found that affect depends in part on
internal attributions, but in a direction Gpposite to Weiner's predictions.
Together, these studies suggest that there may be limitations on applying
the theoretical model proposed by Weiner to real-life situations. However,
this suggestion can only be tentative becauseof the variations in the above
studies with respect to differences in methodology (e.g., assessment of
attributions and affect, populations, nature of "real life" situation).
Weiner, Russell, and Lerman (1978) have suggested that different af-
fects may be associated with different attributions. After reducing an ini-
tial list of 225 emotions to 85, they had subjects read a situational descrip-
tion designed to manipulate attributions and report the affective intensity
they felt the person in the story would feel. Results indicated sets of
affects associated with success or failure, regardless of ascription. Af-
fects such as pleased, happy, satisfied, good, contented, proud, delighted,
and glad were important for success, and uncheerful, displeasure, despair,
and upset for failure. They also reported that particular affects were asso-
ciated with specific descriptions. For success outcomes luck generated feel-
ings of surprise; unstable effort, activation; stable effort, relaxation;
others' effort and personality, self-enhancement; and own personality, self-enhancement. Failure outcomes indicated ability to generate feelings of in-
competence; effort, guilt or shame; personality, resignation; others' effort,
8
4
aggression; and luck, surprise. While this study provides some evidence
that there may be diverse qualitative differences in feelings associated
with particular causal ascriptions, the contrived nature of the research,
in which a subject projects feelings of others, limits generalizability to
real life circumstances.
The research reported in this paper was designed to investigate the
nature of affect associated with causal ascriptions in a real life setting.
There were two primary purposes, 1) to determine the extent of different
affects reported after receiving the results of an actual examination and
2) to assess the relationships of causal attributions to affects. In in-
vestigating whether different affects are reported, a measure of perceived
value was included as exploratory. Although not an emotion, value is an
important evaluative component of attitudes separate from affect (McMillan,
in press), and may be related to causal ascriptions. Also, perceived value
is theoretically important from an expectancy/value model of behavior (e.g.,
Atkinson, 1964; Rotter, 1954). A second exploratory feature was included in
the analyses. In the Weiner et. al. 1978 study, task importance was held
constant in every story; it was always "very important" to receive a high
score. It seems possible, based on research in social psychology concerning
the effect of task importance (e.g., Pittman, Schemer and Wright, 1977),
that this situational variable could have an impact on both the ascriptions
cited and the affect associated with a particular ascription. Thus, per-
ceived importance of doing well on the exam and importance of learning the
material were included in the analyses. It was anticipated, based on pre-
vious research conducted in real life settings, that once the impact of
"success" or "failure" is accounted for, causal ascriptions will contribute
a small amount in predicting affect.
9
6
Method
Subjects
The subjects were students enrolled in two sections of an undergraduateeducational psychology class at the University of Iowa. Both sections wertaught by the same instructor during the 1978 fall semester. The first sec-tion was comprised of 21 males and 48 females (22 sophomores, 28 juniors,and 19 seniors), and the second had 6 males and 35 females (17 sophomores,lb juniors, and 8 seniors).
Procedure
Each section took the same midterm exam during the sixth week of thesemester. On the next class session the students were told that the examswere graded, and they were asked to participate as subjects in a researchproject investigating
"attitudes toward testing." The students completed aninformed consent form and were told that their responses would be anonymousand would have no bearing on course grades. All students agreed to partici-pate and were given Part I of the experiment, a ,six item
questionnaire de-signed to assess the student's
pre-attributions for their outcome - "pre"in the sense that they
were answered prior to handing back the tests. Thequestions assessed the following using a 7-point Likert type scale: effort(Much more than usual -- about the same as use, much less than usual);how well the student usually did in the area (very well -- O.K. -- verypoorly); importance of doing well on the test (very important -- somewhatimportant); test difficulty (very difficult -- very easy); and confidencethat they dia
(very confident -- not confident).After completing
these questions, the tests were returned to the students, discussed, and adistribution of scores with grades was presented. The raw scores rangedfrom 25 to 50, the mean was 40, and the standard deviation was 6. For both
10
sections together there wore 14 As, :!) 1/ t. Os, and 9 is given
to the students fv, urddds, Ndd the yhd of the idihy session students were
Asked to respond (.0 A few mere quesiloos (Pdrts 11 dnd 111), Part t I Wch
d listing of 1t affects on do 8-point scale (anchored with the terms not
di dud "waromulY"). and students indicated their reeling% dhow: the
test by making a check on the scale. The affects Included those Weiner et,
al. (1978) had found as either particularly salient fur success ur failure
outcomes, (e.g., pleased, happy, shame) and those which were discriminating
(e.g., competence, surprise) for the four most common causal ascriptions
(effort, ability, task difficulty, and luck). The 15th word, valuable, was
included in the list of affects as exploratory. In Part III the students
were asked to indicate the extent to which they considered their performance
a success or failure by checking the appropriate area on a 14-point scale
(extreme success, not much of a success or failure, extreme failure), and
were asked to indicate to what degree (by percentage adding to 100%) each
of the four causal ascriptions was responsible for their performance.
Finally, students were asked to indicate the test score range their score
fell in on the questionnaire to enable the investigators to identify "objec-
tive" performance as well as perceived performance. The students attached
Part I to Parts II and III and returned them to the instructor, and were
then completely debriefed about the study.
Results
The intercorrelations of the 15 affect variables were examined to
investigate whether separate affects were reported. Several correlated
at .90 and higher. Further, each of the 15 affects was not considered to
represent a unique aspect of the affective domain. Thus, a data reduction
11
w4 htiplumehtod A 0 int Ipal Lomphhoht 0114Iv%I a waa 110101Uott On
the 14 Affett. yall4titc4, it to 1141 vhidow.atory 14010'0 0 the Ihth word
(valuable), 11 was not iptilocu in data rednclion and WA!), CO11
00drUtl da d separate criterion oittouve. Ihe an resulted in the crea
tion X11 tut) hrthoThuil tompohqht-., Iahulvd I, I ohd uoch a t:omhlwaioh
All 14 affect variables, !able I illustrates the rifted loadings of the
components, Cl sows to the presence of a singular positive emotional
i'dblo
response, in which the student feels happy, competent, proud, etc. C2
reveals a feeling of surprise, a reaction to an unexpected outcome which
could be either positive or negative. Cl and C2 component scores were com-
puted for each student. These scores, plus each student's score on the
"valuable" affect variable (Ca) were the criterion measures in subsequent
analyses, rather than analyzing each affect separately.
A 2 x 2 MANOVA was conducted to determine whether the student's class
section or sex should be taken into account in any further data analyses.
Results indicated no significant section or sex effect.
Stepwise multiple regression techniques were used to determine the
extent to which each student's 10 attributional responses regarding test
performance predicted affects generated after receiving test results (6
responses before receiving test results, 4 after). Two additional indepen-
dent variables, the student's perceived success or failure on the test (PSF),
and the actual score range the student's test score fell within (SR), were
forced into all regression equations first since the question of interest
concerned how much pre- and post-test attributions aid in predicting affects
once actual outcome variables (assumed to be assessed by PSF and SR) are
12
8
taken into consideration. Table 2 indicates the results of the regression
analyses for Cl, C2, and C3.
Table 2
For Cl (happy, proud, content, etc.) only PAT 3 (post-attribution to
task difficulty) and X1 (pre-attribution to effort) were found to add sig-
nificantly once PSF and SR were taken into consideration. For C2 (surprise),
PSF and SR were not very powerful predictors, but again only 2 of the attri-
butional variables were found to add significantly, PAT 2 and X3 (post-attri-
bution to effort and pre-attribution to importance of doing well). Finally,
only PAT3 was found to aid in predicting C3 (valuable) once PSF and SR were
taken into consideration.
An examination of the regression coefftctents for the Cl regression
equation suggests that students' feelings of competence, happiness, etc.,
are positively related to the amount of effort they report exerting in studying
for the test and negatively related to the task difficulty. Feelings represented
by C2. (surprise) are positively related to the degree to which the students
felt effort was responsible for their performance, and positively related to
the importance of doing well. Reported value (C3) was negatively related to
task difficulty. Thus, students felt happiest when they report high effort in
studying for the test and when the percentage of outcome explained by task
difficulty is low. As importance of the task and post-attribution to effort
increased so did surprise, and value increased as the percentage of the
outcome due to task difficulty decreased.
A second analysis was done which considered all four post-attributions
simultaneously as an attributional pattern. Subjects were grouped in a post
hoc fashion on the basis of their post-attribution profiles, and additional
analyses were done on the basis of these profiles. Four profiles were identified,
13
9
and students were classified into 1 of 4 groups on the basis of these 4
profiles. Students whose profiles did not match any of the 4 were not
considered in subsegLient analyses. Group 1 consisted of 28 students, all
of whom had high post-attributions on effort with low attributions on all
three other factors. Group 2 consisted of 14 subjects who had high post-
attributions to ability with low attributions on all three other factors.
Group 3 consisted of 11 subjects who had high post-attributions to task
difficulty with low attribution on all three other factors. Finally,
group 4 consisted of 15 subjects who had high attributions to both ability
and effort, but low attributions on task difficulty and luck. Representative
profiles for each group are illustrated in Graph 1.
Graph 1
A one way MANCOVA was used to determine if the affects reported by
students in one profile were significantly different from affects of students
in other profiles. Perceived success and failure (PSF) and score range (SR)
were used as covariates to control for the effect of outcome on affects. The
MANCOVA was significant (p <.04), and univariate ANCOVAs for each affect are
reported in Table 3. The means and variances of each group's affects,
adjusted for the effects of PSF and SR, are reported in Table 4. The results
indicate that the groups are significantly different for C3 (value), but
not Cl (happiness) or C2 (surprise). An analysis of the cell means and
the direction of the regression coefficients indicates that students who
attribute their outcome primarily to task difficulty report they value the
test less than students who attribute the outcome to effort or ability. No
significant differences between groups 1, 2, or 4 (effort, ability and
effort/ability, respectively) are observed for C3. The ANCOVA for Cl
14
10
approaches significance, (p<.10) and the cell means indicate that students
wno attribute their outcome primarily to effort report more happiness,
pride, etc., than students who report high ability or task difficulty at-
tributions. However, given the influence of perceived success and failure,
(accounting for 71% of the variance), the relationship of the effort attri-
bution to affect is small.
Tables 3 & 4
Discussion
The results of the principal component analyses indicates, for these
students, the presence of two distinct affects.' There was no evidence of a
meaningful differentiation between enjoyment, contentment, pride, etc.
Rather, these affects loaded very high with a single component. The affects
of surprise and astonishment were distinct from those comprising Cl and
together represented the second component. The exploratory factor, "valuable"
did not correlate very highly with other affects (.2 to .4) and seemed to
represent a distinct reaction which deserves further study. However, there
are limitations to the suggested interpretation of the "affect" results.
First, only 15 affects were assessed, and while they represented the most
common affects reported by students in earlier research by Weiner, the in-
clusion of additional affects might generate other distinct components.
For instance, affects of "relaxation," "activation," "gratitude," and "resig-
nation" might establish other components. Second, a different form of meas-
urement of the affects might give different results. This study used a
continuum from "not at all" to "extremely" for each affect. Using adjective
I 5
11
pairs (enjoy-hate) or free responses might provide different results since
"not at all" doesn't always imply the opposite of a positive affect (i.e.,
delighted-disgusted might be answered differently than extremely delighted-
not at all delighted).
The results of the regression analyses for Cl, C2, and C3 using both pre-
and post-attributions suggests that perceived success/failure is clearly
the most important factor related to affect. For Cl once perceived success/
failure was entered into the analysis, the remaining variables added only
5% to the total variance accounted for. This finding is consistent with
other studies in real-life situations (Bailey, Helm & Gladstone, 1975;
Arkin & Maruyama, 1979; Frieze, Snyder and Fontaine, 1977). As would be
predicted on the basis of previous research (Weiner & Kukla, 1), happiness,
pride, etc. are greater as external causal attributions become smaller (in
this case the contribution of task difficulty), but there was no evidence
of a positive relationship between Cl and internal post-attributions. The
pre-attribution of degree of effort exerted in studying for the test was
positively related to Cl and since this was a perception reported before
students knew their test scores, it is not technically a causal factor. It
may represent, then, a condition which influences affects, regardless of
post-test causal attributions.
The regression analyses for C2 suggests that a small positive relation-
ship exists between the affect of surprise and post-attributions to effort.
This relationship may indicate that as effort becomes a more important attri-
bution in explaining performance, surprise increases. This result is some-
what contradictory to other research which indicated surprise is primarily
related to luck (Weiner et. al., 1978). It seems reasonable, however, that
students who scored high on the test and were pleasantly surprised would
16
12
report that effort, an unstable internal attribution, was also high. That
is, their perception of effort as contributing to their score could increase
after receiving a surprisingly high test score. Students who scored low
as expected, on the other hand, would be more likely to report that effort
was not an important causal factor, since high effort/low score would suggest
that the student did not have the ability to do well. The negative relation-
ship of importance of doing well and surprise suggests that students who re-
ported that performance was not important and received a high score would
be pleasantly surprised.
The negative relationship between C3 (valuable) and the task difficulty
post-attribution implies that value increases as the external attribution of
task difficulty decreases. This result is consistent in direction with Cl
and with previous research, though "valuable" has not been included as a
criterion measure in other studies. The analyses may have failed to pick up
the opposite relationship for internal attributions due to the fact that
post-attributions were assessed in an ipsative fashion, which resulted in
luck being given almost no percentage for causality. Thus, the internal at-
tributions were split between ability and effort, while task difficulty was
the only external attribution which received a significant part of the 100%
total. However, the single most important impression that is evident from
all three regression analyses, despite the difficulty in using ipsative meas-
ures, is the relatively small contribution both pre- and post-attributions
make in explaining affective reactions.
The MANCOVA done to investigate whether reported affects were different
for the four profiles indicated that value was higher for profiles in which
internal attributions were higher than attributions to task difficulty. No
other statistically significant results were attained but a similar pattern
13
of results approach significance for Cl (happiness, pride, etc.). The
difference between the adjusted profile group cell means for Cl also indi-
cates, however, that students who attributed the outcome primarily to
effort (group 1) reported greater happiness, pride, etc. than those who
reported ability as the primary attribution (group 2). These results are
also consistent with previous research (Weiner & Kukla, 1970) which suggests
that internal attributions maximize positive affect and external attribu-
tions maximize negative affect.
The results of this study provide several implications for attribution/
affect questions. First, it seems that students may react affectively in
a generalized, global manner immediately following knowledge of test results.
The major limitation to this suggestion is that results attained may be
unique for the subjects used, college students, and the nature of the data
analyses, which obscures individuals who may have distinct affective reactions.
That is, there may not have been sufficient individual differences among the
students to capture distinct emotions in the principal component analyses.
If this was true, then the results represent the first level of Weiner's
(1979) recent conceptualization of how affect is related to attributions.
He suggests that, based on results of college students' reactions to par-
ticular scenarios, following this generalized reaction a number of discrete
affects are generated, based on different attributions. It seems that the
only way to test Weiner's hypotheses is to research these relationships
further in real life situations since his assumes that such scenarios actually
occur. Considering all individuals together, though, this generalized af-
fective reaction seems clear. Second, it seems that perceptions of value
are distinct from this generalized affect and may be significantly related
to attributions. It is both interesting and encouraging that value is not
.1..8
14
predicted very well by success or failure outcomes. The results also
suggest that task importance may be related to perceptions of value.
Third, task outcome and perceptions of success or failure seem to be of
primary importance in predicting affect. A recent article by Weiner (1979)
suggests that the importance of task outcome is major but that over time
the attributions made for success or failure will have a greater impact on
the individual's self-esteem. This hypothesis was not tested in this
study, and it may be that the relatively small impact of attributions
found may be partially a function of. the immediacy of the assessment.
While the lack of much of a relationship between affect and attribution
could be explained by Weiner's (1979) reconceptualization stated above,
Covington and Omelich (1979) provide an alternative hypothesis. They sug-
ge.,..-t t)at attributions are reactions to rather than causes of achievement,
and as postdictive reactions attributions are prone to reflect self-serving
biases. As such, they obscure their causal role as information-processing
agents. Thus, individuals may form attributions, depending on an outcome,
which maintains a high sense of self-worth. The presence of such self-
serving rationalizations may obscure cognitively based attributional linkages.
A third hypothesis seems equally plausible, that in real life individuals
simply don't process complex attribution-affect relationships. In most cir-
cumstances, individuals may not mentally construct a distinct attribution
or distinct affects. The reaction, rather, is global and simple. Further
research is needed to investigate this question to determine the extent to
which children at different ages meaningfully process various attributions
and affects.
Fourth, the results of this study suggests that while outcome is the
most important factor in determining reactions, internal attributions are
.1.9
15
related to both affect and value. Fifth, the profiles generated from the
ipsative measures of attributions provides an alternative way to concept-
ualize causal attributions. The profiles represent patterns which represent
more precisely the relative contribution of each attribution. Since causal
factors other than the four used in this study could also be important in
certain situations, and since the patterns resulting from these studies are
based on a relatively small number of subjects, further research of how
attributions can be represented as integrative patterns rather than singular
factors is needed. It would be useful, for instance, to investigaU whether
attributional patterns could be used in more controlled, experi;neritai re-
search as well as in an ex-post facto manner as is demonstrated in this
study.
ti
16
References
Arkin, R. M. and Maruyama, G. M. Attribution, affect, and college exam
performance. Journal of Educational Psychology, 1979, 71, 85-93.
Atkinson, J. W. An Introduction to Motivation. Princeton, N.J.: Van
Nostrand, 1964.
Bailey, R. C., Helm, B. and Gladstone, R. The effects of success and failure
in a real-life setting: Performance, attribution, affect, and expect-
ancy. Journal of Psychology, 1975, 89, 137-147.
Bar-Tal, D., and Darom, E. Pupils' attributions of success and failure.
Child Development, 1979, 50, 264-267.
Covington, M. V., and Omelich, C. L. Are causal attributions causal?
A path analysis of the cognitive model of achievement motivation.
Journal of Personality and Social Psychology, 1979, 37, 1487-1504.
Frieze, I. H., Snyder, H. N., and Fontaine, C. M. Student attributions and
the attribution model during an actual examination. Paper presented
at the Annual Meeting of the Americal Psychological Association,
San Francisco, California, 1977.
McMillan, J. Attitude Development and Measurement. In J. McMillan (Ed.),
The Social Psychology of School Learning. New York: Academic Press,
in press.
Nicholls, J. G. Effort is virtuous, but it's better to have ability:
Evaluative responses to perceptions of effort and ability.
Journal of Research in Personality, 1976, 10, 306-315.
Pittman, T. S., Scherrer, F.. W., and Wright, J. B. The effects of commit-
ment on information utilization in the attribution process. Personality
and Social Psychology Bulletin, 1977, 3, 276-279.
17
Reimer, B. S. Influence of causal beliefs on affect and expectancy.
Journal of Personality and Social Psychology, 1975, 31, 1163-1167.
Rotter, J. B. Social Learning and Clinical Psychology. Englewood Cliffs,
N.J.: Prentice-Hall, 1954.
Sohn, D. Affect-generated powers of effort and ability self attributions
of academic success and failure. Journal of Educational Psychology,
1977, 69, 500-505.
Weiner, B. Achievement motivation as conceptualized by an attribution
theorist. In B. Weiner (Ed.), Achievement Motivation and Attribution
Theory. Morristown, N.J.: General Learning Press, 1974.
Weiner, B. Attribution and affect: Comments on Sohn's critique.
Journal of Educational Psychology_, 1977, 69, 506-511.
Weiner, B. A theory of motivation for some classroom experiences.
Journal of Educational Psychology, 1979, 71, 3-25.
Weiner, B., and Kukla, A. An attributional analysis of achievement motiva-
tion. Journal of Personality and Social Psychology, 1970, 15, 1-20.
Weiner, B., Frieze, I. H., Kukla A., Reed, L., Rest, S., and Rosenbaum, R. M.
Perceiving the Causes of Success and Failure. Morristown, N.J.:
General Learning Press, 1971.
Weiner, B., Russell, D., and Lerman, D. Affective consequences of causal
ascription. In J. H. Harvey, W. J. Ickes, and R. F. Kidd (Eds.),
New Directions in Attribution Research, (Vol. 2). Hillside, N.J.:
Erlbaum, 1978.
18
Table 1
Principal component analysis factor loadings for affects.
Cl C2
surprised -0.12690 0.70818happy 0.95296 0.07937competent 0.86512 0.04030satisfied 0.94198 0.08171hopeless -0.62894 0.42377delighted 0.92932 0.18855proud 0.94285 0.06104Pleased 0.96246 0.09576inadequate -0.77675 0.31409good 0.90750 0.05707contented 0.91503 0.12729upset -0.77150 0.32659astonished -0.23521 0.80753elated 0.68672 0.52651
Eigenvalue 9.0208 1.8982
Percent varianceaccounted for 64.4 13.16
19
Table 2
Summary of Stepwise Multiple Regression for Cl, C2, and C31'2
Cl (competent, happy, proud, etc.)
Source df MS F P R2
regressioncoefficients
RegressionPSF,SR
PAT3X1
Error
4
2
1
1
83
16.8831.522.641.84
.25
68.73128.3510.737.50
4.0014.001.01<..05
.77
.72
.03
.02
-.50-.37-.01
`C2 (surprised)
Source df MS F P R2
regressioncoefficients
RegressionPSF,SR
PAT2X3
error
4
2
1
1
83
3.662.386.823.06
.87
4.222.747.863.53
<.01
<.05
.17
.05
.08
.04
.30
.01
-.25
C3 (value)
Source df MS F P R2 reression
coefficients
Regression 3 18.18 10.19 (.001 .27PSF,SR 2 15.95 8.94 <.001 .16 -.15PAT3 1 22.65 12.70 <.001 .11 -.04
error 83 1.78
1. PSF and SR were forced into each equation first.
2. High values of PSF were keyed to indicate failure; high values of SR werekeyed to indicate low scores; high values of PAT 3 were keyed to indicate highattribution to task difficulty; high values of X1 and X3 were keyed to indicatelow effort and low importance, respectively.
2 4
20
TABLE 3
Analysis of Covariance of Affects for Profile Groups
Cl
Source
RegressionPSF, SRGroup/PSF, SRError
df
5
2
3
59
MS
10.1824.47
.66
.305
C2
F P
33.41 '.00580.23 '.0052.16
R regressionefficients
.739
.710 -.59
.029 -.46
Source df MS F P R2 regresiioncoefficients
Regression 5 .670 .70 .056PSF, SR 2 .813 .855 .027 .14Group/PSF, SR 3 .572 .601 .029 .96Error 59 .951
C3
Source df MS F P R2 regressioncoefficients
Regression 5 7.28 4.30 '.005 .267PSF SR 2 8.69 5.14 <.01 .127 -.26Group/PSF SR 3 6.34 3.75 (.05 .140 -.61Error 59 1.69
21
TABLE 4
Means and standard errorsof profile groups for Cl, C2 and C3
adjusted for the effects of PSF and SR
1
Profile Group
42 3
.25 -.10 -.23 .10Cl Se .11 .16 .77 .14
n 28 14 11 15
.06 -.14 -.03 -.36C2 Se .19 .27 .31 .25
n 28 14 11 15
R 4.81 4.88 3.24 4.52
C3 Se .26 .37 .42 .34
n 27 13 12 15
4> 9
100 10090 9080 8070 7060 6050 5040 4030 3020 2010 1000 00
PAT1 PAT2 PAT3 PAT4Ability Effort Task Luck
GROUP 1N = 28
100 10090 9080 8070 7060 6050 5040 4030 3020 2010 1000 00
PAT1 PAT2 PAT3 PAT4
GROUP 3N = 11
22
PAT1 PAT2 PAT3 PAT4
GROUP 2N = 14
PAT 1 PAT2 PAT3 PAT4
GROUP 4N= 1.5
PAT 1 PAT 2 PAT 3 PAT 4(Ability) (Effort) (Task Difficulty) (Luck)
Group 1 15.72
( 9.35)*
67.69(12.00)
12.48( 9.66) (
4.104.97)
Group 2 55.57 23.57 14.64 6.21(14.93) ( 7.70) (11.00) ( 7.01)
Group 3 19.17 24.17 50.00 6.67
( 9.50) ( 7.70) ( 7.39) ( 8.62)
Group 4 38.60 41.33 12.67 7.40( 3.89) ( 5.16) ( 5.30) ( 4.47)
Graph 1. Profiles, means and standard deviations 1of
four post-attribution patterns.1The mean number indicates percentage attributed to each factor.
*Standard Deviation in parentheses.